Experimental data [on fracture]

To investigate the model in the main text requires looking not only at the path of a crack, but also at dislocations of atoms near it. To do this dynamically is difficult, but in a perfect crystal final patterns of dislocations that remain at the edge of a region affected by fracture can be seen for example by electron diffraction. And it turns out that these often look remarkably like patterns made by 1D class 3 cellular automata. (Similar patterns may perhaps also be seen in recent detailed simulations of fracture processes in arrays of idealized atoms.)